Stick-slip friction and wear of articular jointsDong Woog Leea, Xavier Banquya, and Jacob N. Israelachvilia,b,1

aDepartment of Chemical Engineering and bMaterials, University of California, Santa Barbara, CA 93106

Contributed by Jacob N. Israelachvili, December 26, 2012 (sent for review October 19, 2012)

Stick-slip friction was observed in articular cartilage under certainloading and sliding conditions and systematically studied. Usingthe Surface Forces Apparatus, we show that stick-slip friction caninduce permanent morphological changes (a change in the rough-ness indicative of wear/damage) in cartilage surfaces, even undermild loading and sliding conditions. The different load and speedregimes can be represented by friction maps—separating regimesof smooth and stick-slip sliding; damage generally occurs withinthe stick-slip regimes. Prolonged exposure of cartilage surfaces tostick-slip sliding resulted in a significant increase of surface rough-ness, indicative of severe morphological changes of the cartilagesuperficial zone. To further investigate the factors that are condu-cive to stick-slip and wear, we selectively digested essential com-ponents of cartilage: type II collagen, hyaluronic acid (HA), andglycosaminoglycans (GAGs). Compared with the normal cartilage,HA and GAG digestions modified the stick-slip behavior and in-creased surface roughness (wear) during sliding, whereas collagendigestion decreased the surface roughness. Importantly, frictionforces increased up to 2, 10, and 5 times after HA, GAGs, andcollagen digestion, respectively. Also, each digestion altered thefriction map in different ways. Our results show that (i) wear is notdirectly related to the friction coefficient but (ii) more directly re-lated to stick-slip sliding, even when present at small amplitudes,and that (iii) the different molecular components of joints worksynergistically to prevent wear. Our results also suggest potentialnoninvasive diagnostic tools for sensing stick-slip in joints.

arthritis | biolubrication | biointerface | boundary lubrication

Osteoarthritis (OA) is widely present in the elderly (1, 2) aswell as people with obesity (3). Despite the prevalence of

OA, the physical and biological events that trigger articularcartilage damage are still poorly understood, especially at themolecular and submicron levels; so far, no model has beensuccessful in describing the mechanistic cascade that leads tocartilage wear and irreversible loss (4). There are no availabletechniques to detect the onset of joint diseases, and tracking theprogression of these diseases remains a major challenge (5).One of the key challenges for establishing a model that can

successfully track the cause and progression of joint diseases is inthe fact that articular joints are designed to function under anextremely large range of mechanical, dynamic, and environ-mental conditions, under any of which a priori damage can oc-cur. Many scientists have, therefore, focused their efforts onspecific physiological regimes under which joints would be se-verely subjected to mechanical damage caused by a failure of thelubrication mechanism at the submicron level.The most important and also least understood physiological

regime associated with damage is referred to as the boundarylubrication (BL) regime, where two sliding surfaces and themolecular species that are attached to them are in intimatecontact during most of the motion. It is now well-known that, inthe BL regime, wear of soft surfaces, especially cartilage, caneasily and quickly propagate as long as the surfaces are subjectedto a high enough normal load (typically above several Newtonscorresponding to normal pressures ranging from 1 to 10 MPa)and sliding speeds (typically above millimeters per second) (6).Under these conditions, the rheological (viscous) properties ofthe lubricating fluid do not contribute significantly to the lubri-

cation of the surfaces, and it is the macromolecules that arepresent in the interstitial gap separating the surfaces that lubri-cate and ultimately, protect the surfaces from wear. Boundarylubrication can be achieved in many ways and is not limited tohigh loads or prolonged sliding; it can also be achieved undermild loading conditions (typically by applying a load in the rangeof several milliNewtons corresponding to a normal pressure of∼0.1 MPa) as long as the sliding speed is low enough (or zerounder quasistatic conditions such as sitting or standing) to avoidelastohydrodynamic deformations of the surfaces and the de-velopment of a thick lubricating fluid between the surfaces. Wearof articular surfaces has never been studied under such mildconditions, and therefore, no mechanism has been proposed todescribe its occurrence.In this study, we show that, under mild sliding conditions cor-

responding to conditions of a human being at rest, stable inter-mittent sliding between two articular surfaces can be observed.Together with stiction spike (7, 8), stick-slip sliding is commonlyat the origin of irreversible transformations of surfaces and inthis study, cartilage topography—a hallmark of all cartilage wearprocesses. By selectively digesting specific components [collagen,hyaluronic acid (HA), and glycosaminoglycans (GAGs)] of thecartilage surfaces, we aimed to establish correlations betweenstructural changes of joint components and their subsequentimpact on the frictional map of the articular joint as they occurduring OA.Many different experimental setups have been proposed for

measuring frictional forces between cartilage surfaces (9–12).Most of these setups can be schematically represented by a me-chanical equivalent depicted in Fig. 1A. When characterizingcomplex frictional behaviors like stick-slip, one can readily seethat the measured average friction force is not the only impor-tant or even the most important parameter defining the systemwhen intermittent friction is present (including other transienteffects, such as stiction). The inertia of the moving parts, thesystem resonance frequencies, the mass (m), and the stiffness (K)are also key experimental parameters.

Significance

The goal of this study was to use the Surface Forces Apparatusto examine the effects of slip-stick friction on cartilage surfacemorphology under different loading and sliding conditions.Different load and speed regimes were represented usingfriction maps that separated regimes of smooth and stick-slipsliding. The finding of this work is that damage generallyoccurs within the stick-slip regimes and is not directly related tothe friction coefficient. Prolonged exposure of cartilage surfa-ces to stick-slip sliding resulted in a significant increase ofsurface roughness, indicative of severe morphological changes(damage) of the cartilage surfaces.

Author contributions: D.W.L., X.B., and J.N.I. designed research; D.W.L. and X.B.performed research; D.W.L. analyzed data; and D.W.L., X.B., and J.N.I. wrote the paper.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. E-mail: jacob@engineering.ucsb.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1222470110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1222470110 PNAS | Published online January 28, 2013 | E567–E574

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

14,

202

0

As can be seen in Fig. 1A, the applied force f by the springcoupling the drive to the stage and the speed v of the drive aredifferent from the true interfacial friction force, f0, and slidingspeed, vint. Starting from rest, the stage will move when the ap-plied force f reaches the friction force, f0. If the static friction(also known as the stiction) force, fst, is greater than the kineticfriction force, fk (Fig. 1C), the stage will accelerate rapidly toa speed, vint, much greater than the slider speed, v, and thecoupling spring will be compressed. At this point, the stage willdecelerate to a stop, where the friction force recovers to its highvalue, fst. This cycle, called stick-slip, can be repeated periodi-cally as long as the driver moves at a constant velocity.Different models have been developed that aim to establish

how macroscopic system properties and nanoscopic (i.e., mo-lecular ordering) properties cooperate to create complex fric-tional motions like stick-slip. For a review of the differentmechanisms of stick-slip, the reader is referred to refs. 13 and 14.Our experiments show that the stick-slip motion of cartilagesurfaces is clearly velocity-dependent and restricted to a velocitywindow that depends on the applied normal load, L. In general,velocity windows for stick-slip are determined by two criticalvelocities (15): the first velocity, vIc, corresponds to the appear-ance of stick-slip and the onset of the velocity weakening regimeof the friction force, f0. The second velocity, vIIc , corresponds tothe disappearance of stick-slip, at which point f = fk = constant.In between vIc and vIIc , different types of stick-slip motions can beobserved, ranging from regular stick-slip (Fig. 1A) to irregularstick-slip (both in amplitude and frequencies, and stiction (Fig. 1D and E). The determination of vIc and vIIc as a function of theapplied load allows a friction map to be established that char-acterizes the frictional behavior of a system.

Results and DiscussionA series of friction force measurements with cartilage againstcartilage (Fig. 2A) were performed over a wide range of drivingspeeds (0.03–100 μm/s) and loads (5–250 mN) (details in Mate-rials and Methods). Morphological studies were also performedby monitoring the cartilage surfaces before and after sliding us-ing interferometry. Additionally, cartilage surfaces were treatedwith specific enzymes to investigate the effects of essentialcomponents on different friction and lubrication properties. Theessential components of articular cartilage that we selected todigest are HA, GAGs, and type II collagen. All three compo-nents exhibit different and important roles in cartilage lubrica-tion and malfunction: modification or digestion of each com-ponent is known to be highly related to the onset and progressionof OA (11, 16–21).

Normal (Untreated) Cartilage. Fig. 2A shows a schematic of thefriction experiments performed with normal (untreated) carti-lage. Two opposing cartilage surfaces were sheared at a re-ciprocal motion of peak-to-peak amplitude ΔD (∼100 μm) andback and forth driving velocity ±v under the applied load L. TheSurface Forces Apparatus (SFA) chamber was saturated withwater vapor to prevent the evaporation of the liquid dropletbetween two cartilage surfaces during the experiments. [Twohundred microliters equine synovial fluid (Fig. 2 C and D) orPBS buffer solution (all other experiments) were used. Frictionphase diagrams show similar trends regardless of the type of fluidbetween two cartilage surfaces.] Fig. 2B shows the effect of L andv on fk, and two distinct lubrication regimes can be clearly seen:(i) Fluid Film Lubrication (FFL; 0 < L < 10 mN, P ∼ 0.07 MPa)with a low friction coefficient (μ = ∂fk/∂L = 0.01–0.02) and (ii)

A

B

C

E F

D

Fig. 1. (A) Schematic of the principle of SFA used for measuring normal load, L, and friction force, f, between two surfaces in contact with area A under backand forth shearing motion of peak-to-peak amplitude D ∼100 μm and driving velocity, v. The friction force, f, is measured as f = K(Δx − ΔD), where Δx and ΔDare the displacement of the drive and the stage, respectively, and K is the stiffness of the measuring spring. (B) Evolution of friction forces with time measuredbetween two cartilage surfaces in synovial fluid at a driving speed of v = 10 μm/s. Right after loading and applying shear motion, the friction force increasesrapidly and evolves to stick-slip or irregular stick-slip sliding motion (B Insets). After sliding for 3 min, the friction force gradually decreases to a steady-statevalue, and sliding phase becomes smooth. Four distinct sliding behaviors are observed depending on the sliding conditions: (C) regular stick-slip, (D) irregularstick-slip, (E) smooth sliding with stiction, and (F) purely smooth sliding.

E568 | www.pnas.org/cgi/doi/10.1073/pnas.1222470110 Lee et al.

Dow

nloa

ded

by g

uest

on

June

14,

202

0

Diffuse (Osmotic) BL (DBL; L > 25 mN, P ∼ 0.09 MPa) witha much higher friction coefficient (μ = 0.11–0.35) (Fig. S1 showsa conventional Stribeck curve). [This regime is likely the sameas the “Osmotic Lubrication” or “Brush Border Lubrication”regimes previously proposed by McCutchen (22, 23).] The tran-sition load, LT, where the lubrication mechanism changes fromFFL to DBL is LT ∼ 16 mN/m (P ∼ 0.08 MPa), which is similar tothe value measured by Greene et al. (9) (Eq. S1 shows the pres-sure calculation). Additional insights into how v and L influencefriction forces (of normal cartilage) and steady-state sliding pro-files (Fig. 2 C and D) were gained by systematically changing vand L over a wide range of conditions (0.03–60 μm/s and 15–100 mN, respectively).Fig. 2C shows how the friction forces (fst, fs, and fk) change

with v at three different loads (L= 15, 85, and 100 mN), and Fig.2D shows the corresponding friction map. At L = 100 mN, stick-slip was observed at vIc = 0.03 μm/s. As v increases, the magni-tude of the stick-slip spikes, Δf= fs − fk, increases until v= 1 μm/s;fs then decreases until the driving speed reaches v = vIIc ∼ 5 μm/s,where the stick-slip sliding disappears (fs = fk). As the loaddecreases, vIc increases, whereas v

IIc decreases, moving all of the

values closer. Accordingly, the velocity window of stick-slip regimedecreases, and by L = 15 mN, the stick-slip window becomes verynarrow, with only irregular stick-slip being observed.The friction map (Fig. 2D) conveniently illustrates the differ-

ent stick-slip and smooth sliding regimes or windows as a func-tion of L and v—two variables that can easily be controlled. Suchmaps can be used to determine and avoid the critical conditionsthat lead to stick-slip and abrasive wear not only in cartilage butalso, many other lubricating systems.Detailed studies of how shearing alters the morphology of

normal articular cartilage were carried out by interferometricimaging because of the ability to scan areas and fast acquisitiontimes, which has been used to image cartilage surfaces (24)(Materials and Methods). Approximately 1-μm-high and approx-imately 1-μm-deep valleys were observed on normal cartilagesurfaces with an rms height roughness of Rq = 340 ± 20 nm.When sheared for 1.5 h (1 h under stick-slip sliding and 0.5 h

under smooth sliding conditions), there was no change in theroughness. However, after sheared for ∼10 h under stick-slipconditions, a significant and systematic increase in the rough-ness (Rq → 460 ± 40 nm) was observed, whereas no morpho-logical change was observed under ∼10 h shearing undersmooth sliding conditions (Fig. S2). These results imply that,even under mild loading and sliding conditions, healthy carti-lage can suffer irreversible damage but only under prolongedstick-slip sliding conditions.

HA-Digested Cartilage. HA is an anionic polysaccharide that iswidely distributed in the human body and especially prevalent inarticular joints. Because of its high molecular mass (3–4 MDa),HA has long been considered to be one of the major (bulk orboundary) lubricants that lowers the friction coefficient andprotects cartilage surfaces from damage (25). Recent studiesshowed that physisorbed HA layers do not have a beneficial ef-fect on lubrication, whereas chemically grafted and/or cross-linked HA layers exhibit excellent wear protection against sur-faces, even when sheared at high pressures (∼20 MPa), regard-less of the high friction coefficient of μ = 0.15–0.3 (26, 27).Experiments involving selective digestion of HA from articularcartilage have shown that HA is mechanically trapped at thecartilage interface by the constricted collagen pore network (9)and that digestion of HA induces a decrease in the porosity andan increase in the stiffness of articular cartilage (10). The molec-ular mass of the HA is known to decrease (from ∼3–4 down to0.5 MDa) with age (18) and/or the progression of arthritis (16, 17).Fig. 3 shows the effects of enzymatic digestion of HA on the

friction forces, friction profiles, and topography of the cartilagesurfaces before and after digestion and shearing. Compared withnormal, healthy cartilage (Fig. 3 A and B), HA-digested cartilage(Fig. 3 C–F) exhibits noticeably different friction characteristicsand topography: (i) a higher friction force (up to two times) (Fig.3C), (ii) a decrease in the second critical velocity and a leftupward shift of the friction map (Fig. 3D), and (iii) significantroughening of the cartilage surface (up to 25%) after 1.5 h of

Fig. 2. (A) Schematic of the experimental setup used with measured and calculated variables. (B) Kinetic friction force (fk) vs. load (L) curve under differentdriving velocities (v = 1.1, 10, and 37 μm/s) showing low-friction coefficient (μ = 0.01–0.2) in the FFL regime (L < 16 mN) and high-friction coefficient (μ = 0.11–0.35) in the DBL regime (L > 16 mN). (C) Friction forces (fst, fs, and fk) vs. driving velocity (v) curve measured at three different loads (L = 100, 85, and 15 mN).The shaded regions indicate the stick-slip sliding regime. (D) Friction map showing representation of cartilage lubrication profiles. The dotted lines indicatethe observed and measured trends based on the experiments and theories (15). (E and F) Topographic images (top view) of the contact zone of normal(nondigested) cartilage (E) before and (F) after 10 h shearing in stick-slip conditions. Red and blue colors indicate higher and lower heights, respectively.Single height profiles are also shown below each image.

Lee et al. PNAS | Published online January 28, 2013 | E569

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

14,

202

0

shearing (1 h under stick-slip sliding and 0.5 h on smooth slidingconditions) (Fig. 3 E and F).The Rq value of HA-digested cartilages increased from 370 ±

20 to 460 ± 40 nm after 1.5 h of shearing (1 h under stick-slipsliding and 0.5 h under smooth sliding), which is similar to theincrease in Rq of normal cartilage sheared under stick-slip con-ditions for 10 h. Significant roughening of the shearing surfacescould be another reason for the observed increase in the frictionforce, although there seems to be no simple or direct correlationbetween the roughness and the friction force (28, 29) for softsurfaces like cartilage.These results show that, at relatively high loads (L > 45 mN,

P > 0.11 MPa), the main function of HA is to act as a surface-anchored protective layer rather than a bulk viscosity lubricant(9). To act as a well-functioning protective layer, HA shouldhave a high enough molecular mass to both be trapped in theporous structure of the cartilage and entangle lubricin (LUB) forefficient boundary lubrication (9, 10). After digestion, the lengthof the HA chains is greatly shortened, and they lose the ability tobecome trapped in the porous structure of the cartilage. Thisincrease in HA chains mobility causes the LUB and HA to bepushed out of the shearing contact under high L and v, leading toan increase in the friction force by a factor of approximatelytwo and changing the stick-slip characteristics.

GAGs-Digested Cartilage. GAGs are known to have a high affinityfor HA, forming large multimolecular aggregates. Such aggre-gates are highly sulfated, forming a large anion immobilizedin the porous collagen network. This entropically confinedGAGs–HA complex increases the osmotic pressure within thenetwork, causing interstitial fluid pressurization and giving rise toimproved load-bearing properties and compressive stiffness of thecartilage (12, 30). Highly hydrated GAGs are also known to en-hance the hydration repulsion between two surfaces, which de-creases the friction coefficient (12, 31–33).In the early stages of OA, aggrecanases are found to be

overexpressed and prevent the formation of GAGs–HA complexby cleaving the G1 domain, which contains the HA binding re-gion (19). When GAGs lose their binding affinity to HA, GAGsrapidly diffuse out of the cartilage matrix. The decrease in GAGsconcentration causes a loss in the ability of the cartilage to inducean osmotic pressure balance within cartilage matrix and increases

its vulnerability to loading and shearing (30). Here, we usedChondroitinase ABC (Case ABC), which destroys mostly thechondroitin sulfate (34) and therefore, has similar effects to thecartilage as aggrecanase.Fig. 4 shows the effects of enzymatic digestion of GAGs on

the friction forces, friction profiles, and topography of cartilagesurfaces before (Fig. 4 A and B) and after (Fig. 4 C–F) Case ABCtreatment. Compared with normal cartilage, cartilage treatedwith Case ABC shows different friction characteristics and to-pography, highlighted as (i) higher friction force (up to 10 times)(Fig. 4C), (ii) increase in vIIc and right downward shift of thefriction map (Fig. 4D), and (iii) slight roughening of the car-tilage surface (up to 10%) after 1.5 h of shearing. It is note-worthy that, although the increase in roughness was much lesscompared with HA-digested cartilage, the friction force increasewas significantly higher.The failure to maintain an osmotic balance within the cartilage

and preserve the hydration layer at the interface explains the sig-nificant increase in the friction force after GAGs digestion. Fur-thermore, interpenetration and entanglements of HA chains be-tween opposing surfaces is facilitated because of the absence ofGAGs, further leading to higher friction forces (35). Regardless, theprotective HA layer is still intact at the interfaces, which minimizesmorphological changes compared with HA-digested cartilage.

Collagen-Digested Cartilage. Type II collagen occupies 60–80% ofthe solid fraction of cartilage to form a porous fibril network (12,36–39) that governs the mechanical properties, deformation re-sponse, and friction of cartilage (30, 40). The disruption of thefibril network and softening of cartilage have been observed atthe onset and progression of OA (20, 21), which could also beinduced by applying type II collagenase (41, 42). Recent studies(10) indicate that enzymatic digestion of collagen causes drasticdisruption of the complex microstructure and stability of the porematrix, jeopardizing the cartilage’s ability to control the transportof interstitial fluid through diffusion and/or flow, and inhibits ef-ficient hydrodynamic lubrication.Fig. 5 shows the effect of type II collagen digestion on the

friction force, friction profiles, and topography of the shearedcartilage surfaces. Compared with normal cartilage (Fig. 4 A andB), collagen-digested cartilage (Fig. 4 C–F) shows differentfriction characteristics and topography, such as (i) higher friction

Fig. 3. (A and C) Friction forces vs. sliding velocity curve and (B and D) friction maps (A and B) before and (C and D) after HA digestion. (E and F) 3D images(top view) of HA-digested cartilage (E) before and (F) after 1.5 h shearing (1 h stick-slip and 0.5 h smooth sliding conditions).

E570 | www.pnas.org/cgi/doi/10.1073/pnas.1222470110 Lee et al.

Dow

nloa

ded

by g

uest

on

June

14,

202

0

force (up to five times) (Fig. 4C), (ii) complete disappearance ofstick-slip friction in the measurable range of L and v (Fig. 4D),(iii) disappearance (flattening, Rq →180 ± 40 nm) of the valleyspresent in healthy cartilage, and (iv) significant abrasive wear,which caused further smoothing (Rq→90 ± 40 nm), after 1.5 h ofshearing. The fact that the friction force increased as the carti-lage surfaces became smoother again indicates that frictionforces and surface roughness are not directly correlated.

Generic Friction Map Under Mild Sliding Conditions and Effect ofSelective Digestion. Fig. 6A shows a generic friction phase dia-gram of articular cartilage determined from experimental dataunder mild sliding conditions at varying L and v. With increasingL (blue to red solid lines), a stick-slip sliding regime seems abovea certain critical load (Lc). With increasing L above Lc, the stick-slip window, as determined by vIc and vIIc , becomes wider, and the

magnitude of the stick-slip spikes [Δf= (fs − fk)] (Fig. 6A, dashedline) increase. When the driving velocity was initially set below vIcand increased to v > vIIc at the fixed load of L > Lc, the cartilagegoes through three lubrication regimes: (i) DBL, (ii) transition(also mixed), and (iii) FFL.The DBL regime occurs when v < vIc. Because of the low ve-

locity, the shearing interface maintains a fluid-like but high-vis-cosity phase, causing smooth sliding. As v increases and exceedsvIc with increasing v, both fk and thin (DBL) film viscosity fall,giving rise to stick-slip behavior (13, 43). When v exceeds thesecond critical velocity, vIIc , the cartilage surfaces are separatedby a thicker or Newtonian liquid film of fluid. The stick-slipbehavior disappears, the surfaces slide smoothly, and fk nowincreases with v. When L is lower than Lc, both critical velocitiesdisappear, exhibiting smooth sliding at all velocities.

Fig. 4. (A and C) Friction forces vs. sliding velocity curve and (B and D) friction maps (A and B) before and (C and D) after GAGs digestion. (E and F) 3D images(top view) of GAGs-digested cartilage (E) before and (F) after 1.5 h shearing (1 h stick-slip and 0.5 h smooth sliding conditions).

Fig. 5. (A and C) Friction forces vs. sliding velocity curve and (B and D) friction maps (A and B) before and (C and D) after collagen digestion. (E and F) 3Dimages (top view) of collagen-digested cartilage (E) before and (F) after 1.5 h shearing.

Lee et al. PNAS | Published online January 28, 2013 | E571

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

14,

202

0

Thus, we conclude that, for cartilage to have stick-slip be-havior, the applied load should be higher than Lc, and the drivingvelocity should be between the two critical velocities, vIc and vIIc .We suspect that there will be an upper critical load where car-tilage becomes too stiff to exhibit stick-slip behavior (13, 44),although we were not able to reach this point.Friction experiments with selectively digested cartilage did

give us a substantial amount of quantitative and qualitative dataregarding the role of the essential components of articular car-tilage (Fig. 6B). Compared with normal cartilage (Fig. 6B, thicksolid line), HA digestion has a minimal effect, slightly shiftingthe generic friction curve to the upper left position (a roughlydoubling of the friction force). Collagen and GAGs digestionshave a more pronounced effect (up to 5 times higher fk forcollagen-digested cartilage and 10 times higher fk for GAGs-digested cartilage). Collagen digestion makes stick-slip frictiondisappear, whereas GAGs digestion shifts the stick-slip regime tohigher v.

Molecular Mechanism of Stick-Slip Friction in Articular Cartilage. Fig.7 presents the schematic of normal and digested cartilage, in-dicating changes in the interfacial and bulk structure of the su-perficial zone after digestion that affect the molecular mechanism

of stick-slip friction. Normal cartilage exhibits the largest stick-slip amplitude, Δf, compared with selectively digested cartilage(Table 1), possibly because of transient adhesive bridges [causedby transient formation of entanglement, interpenetration, and/orspecific polymer–protein interaction, which are dynamically (in-stead of statically) broken] caused by GAG–GAG (45, 46)interactions and/or entanglements between HA–HA (27). Short-period (∼1 h) stick-slip sliding of normal articular cartilage didnot generate morphological changes of the cartilage surface.One reason for this finding is that all of the components innormal cartilage work cooperatively (Fig. 7A), resulting in a lowfriction force and excellent wear protection. GAGs-bound HA iseffectively trapped in the collagen network, and LUB and lipidscollaborate together with immobilized HA for effective lubrica-tion and wear protection at the cartilage interfaces (9). Althoughthe stick-slip peak was relatively larger compared with otherdigested cartilage, normal cartilage had the lowest fk (Fig. 6B)and the best wear protection property, with no observablemorphological changes after shearing for 1.5 h (Table 1).When treated with hyaluronidase, HA chains are cleaved to

a shorter molecular mass. Accordingly, the trapping of HAdecreases, increasing their probability to diffuse out from thecartilage matrix (9). Furthermore, LUB and lipid molecules thatwere bound to HA at the cartilage interface can detach moreeasily together with HA (Fig. 7B), leading to direct collagen–collagen contact (which involves hydrophobic forces) (47), whichis likely to be the origin of the stick-slip friction for HA-digestedcartilage. Direct collagen–collagen contact not only shifts thestick-slip velocity regime but also increases the friction forcesand makes cartilage more susceptible to abrasive wear (Fig. 3).We and others (10, 41) have recently found that HA digestionsignificantly stiffens cartilage, which could have an impact onstick-slip behavior.GAGs-digested cartilage behaves differently compared with

HA-digested cartilage, because cleaving chondroitin sulfate fromthe GAGs side chains allows the negative charge of GAGs todecrease (Fig. 7C). This decrease in negative charge disrupts theosmotic balance between cartilage and the bulk fluid, and thecollagen network loses the ability to hold water effectively insidethe matrix. Because of the absence of the bulky and highly hy-drated GAGs molecules, HA molecules are now more likely tointerpenetrate and entangle across the opposing cartilage sur-faces, which could be a possible origin of stick-slip in GAGs-digested cartilage.Unlike hyaluronidase or Case ABC, collagenase dissects the

collagen fibers (Fig. 7D). The dramatic softening of the collagen

A B

Fig. 6. Generic friction maps of cartilage frictional behavior under mild sliding conditions. (A) Effect of load on friction map. Orange, purple, and gray shadesindicate regular stick-slip, irregular stick-slip, and stiction regimes, respectively. (B) Effect of selective digestions. Dashed lines indicate the extrapolated trendsexpected from numerous experiments and theories (15). Closed circles indicate critical velocities when stick-slip appears (vIc) and disappears (vIIc ). Gray shadesindicate stick-slip regime windows.

HA

CollagenaseChondroitinase ABCHyaluronidase

A Normal

GAGsCartilage

LUB

Lipids

zone

B HA digested C GAGs digested D Collagen digested

Fig. 7. Schematics of cartilage interfaces indicating molecular mechanismof stick-slip friction (A) before and (B–D) after selective digestions.

E572 | www.pnas.org/cgi/doi/10.1073/pnas.1222470110 Lee et al.

Dow

nloa

ded

by g

uest

on

June

14,

202

0

matrix allows for large deformations and rupture of the fibrils tooccur during sliding, which decreases the ability of the surfaces toform transient adhesive bridges responsible for stick-slip motion(Figs. 5, 6B, and 7D).

Significance of Stick-Slip Friction in Cartilage. Stick-slip motion as theorigin of cartilage wear under mild shearing conditions. Many excessivesports and workouts put joint health in jeopardy, because jointsare exposed to extreme conditions, such as high-impact loadsand fast shearing speeds. However, it is surprising that manypeople also suffer from pain when they simply get up from thebed, sit, or stand still for a long time, both of which are known tobe mild conditions for cartilage. These latter symptoms are muchmore prevalent and significant among arthritic patients, but theprimary causes of joint pain under mild conditions are still notwell-known. Stick-slip friction is a very common type of frictionthat is known to lead to abrasive wear (13, 43). We propose thatstick-slip friction in articular cartilage may be one—or even themajor—cause of wear under mild shearing conditions not onlyfor arthritic cartilage but also, healthy cartilage when sheared forprolonged periods of time.Damage caused by stick-slip may also occur under more severe

conditions. Importantly, we find that the stick-slip amplitude Δf(typically in the range 0–2 mN or 0–2 g) or slip length (typicallyin the range 10–50 μm) need not be large for damage to occur;this finding may be the reason that this phenomenon has notbeen previously reported, because most biotribometers measurethe average friction force and usually lack the force, time, anddistance resolution (sensitivity) of the SFA used in these studies.To a noninvasive and cost-effective diagnostic tool for arthritis. In theUnited States, over 50 million adults suffer from arthritis, andthe numbers are growing rapidly. Conventional diagnostic toolsfor arthritis are often invasive (e.g., synovial fluid test and bloodtest) and/or costly (e.g., X-ray and MRI), and the demand fornoninvasive and cost-effective diagnostic tools for arthritis isincreasing. Although these clinical in vivo tests should, of course,be performed, our results indicate that more sensitive frictiontests should be implemented that could identify early stages ofcartilage degeneration. Stick-slip friction can be recorded by thesound that it generates (compare with the sound from a violin,the squeaking noise of a door, the noise from a breaking car,etc.) (43). An acoustic device could easily detect and systemati-cally analyze the stick-slip characteristics (frequency and ampli-tude) of a patient and could be extremely useful as a cost-effective and noninvasive screening tool for arthritis before or inconjunction with other routine diagnoses and treatments.

Materials and MethodsCartilage Sample Preparation. Porcine leg was purchased from Sierra forMedical Science, and cartilage tissue was dissected from porcine knee articularjoints no later than 1 d after slaughter. Dissection was performed under roomtemperature in a laminar hood to minimize contamination. During dissection,cartilage sampleswere kept in awet condition by continuous rinsingwithHBSS.Full-thickness (∼3 mm, from the superficial zone down to the calcified zone)cartilage was dissected with a razor blade and stored in Hanks’ solution at−50 °C until use (storage duration was at most 1 mo). Two flat glass disks werecleaned using chloroform followed by ethanol before use. The flat sides of thecartilage samples were partially dried using lint-free cloth and then glued ontoglass disks using poly(cyanoacrylate) adhesive followed by curing in PBS bufferfor 30 min.

SFA Experiment. Glued cartilage surfaces were mounted to an SFA2000equipped with a friction device or a 3D sensor/actuator with bimorph slider(48). In this study, an SFA2000 is used as a microtribometer (49) with highresolution in measuring both load and friction (up to ∼0.01 mN). The SFAchamber was saturated with water vapor to minimize evaporation of waterin cartilage. Between the two facing cartilage surfaces, 200 μL equine sy-novial fluid (Alamo Pintato) for the experiment in Fig. 2 C and D or PBS(Sigma) for all other experiments were injected as a reservoir. Loading andshearing were conducted with course-control micrometer and bimorph slider,respectively, whereas the friction forces and loads were measured usinga recently developed 3D sensor/actuator (48, 50). Because of the opacity ofthe samples, optical interferometry could not be used as in standard SFA-Fringes of Equal Chromatic Order (FECO) experiments. Although the cartilagesurfaces were being compressed under a constant load, they were shearedlaterally at a fixed driving velocity ranging from 0.03 to 100 μm/s with slidinga amplitude of ∼100 μm until the measured loads and friction forces reachedthe steady state (Fig. 1B). After measuring friction forces at a wide range ofdriving velocities, the load was varied, and friction forces were again mea-sured. Measurements were repeated for three to four different loads. AfterSFA experiment with normal cartilage, 50 μL enzyme for selective digestion(hyaluronidase, 200 units/mg in PBS; Sigma), type II collagenase (2,000 units/mg in PBS; Sigma), and Case ABC (137 units/mg in PBS; Sigma) were injectedbetween the surfaces and equilibrated for 3 h. After digestion, surfaces wererinsed with PBS buffer. The SFA experiments were then repeated withdigested cartilage using the same procedure as mentioned above.

Interferometric Imaging. A Wyko 1100 white-light interferometer (Veeco)was used to image surface topography of the cartilage. Right beforemounting the sample on the stage, the cartilage surface was briefly driedwith nitrogen under a laminar flow hood to minimize abnormal signalsgenerated by water reflection. Cartilage position was carefully adjusted tothe center of the stage to image the highest point, which is actually beingsheared in the SFA experiments. The rms roughness Rq was calculated byaveraging fiv random windows (80 × 80 μm) after tilt and curvature fittingusing Vision32 software.

Glued cartilage surfaces were digested with each enzyme and imagedwitha Wyko 1100 white-light interferometer before shearing to characterize theeffects of digestions. After imaging, they were sheared for 1.5 h (1 h understick-slip sliding and 0.5 h under smooth sliding) and imaged again to in-

Table 1. Comparison of stick-slip and wear between normal and selectively digested cartilages

Normalized stick-slip Δf/fsMorphological change after 1.5 h

(1 h stick-slip/0.5 h smooth) shearing Reasons

Normal High None* Interpenetration and effective trapping of the HALubricin can bind/bridge both surfacesGAGs–GAGs interaction

HA-digested Medium Medium Less HA trapped in pore structureLess stability of the HA-bound LubricinDirect collagen–collagen contact

GAGs-digested Low Low Interpenetration/bridging of the HALubricin can bind/bridge both surfaces

Collagen-digested None High Deformations and rupture of the collagen fibrils

*When sheared for 10 h under stick-slip conditions, it showed medium change of surface morphology.

Lee et al. PNAS | Published online January 28, 2013 | E573

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

14,

202

0

vestigate the effects of the shearing. For normal (nontreated) cartilage,imaging was performed under the same procedure; however, because it didnot show any change in roughness after 1.5 h of shearing, 10 h of shearingunder stick-slip condition were performed to see the effects of prolongedstick-slip shearing on healthy cartilage.

ACKNOWLEDGMENTS. We thank Alamo Pintado Equine Medical Center forsupplying us with equine synovial fluid. We also thank Drs. David Bothmanand Kimberly Turner for access to the Wyko Interferometer. X.B. would liketo thank the Otis Williams Fellowship for their support. This work wasfunded by McCutchen Foundation.

1. McAlindon TE, Wilson PWF, Aliabadi P, Weissman B, Felson DT (1999) Level of physicalactivity and the risk of radiographic and symptomatic knee osteoarthritis in theelderly: The Framingham study. Am J Med 106(2):151–157.

2. Felson DT, et al. (2000) Osteoarthritis: New insights. Part 1: The disease and its riskfactors. Ann Intern Med 133(8):635–646.

3. Grotle M, Hagen KB, Natvig B, Dahl FA, Kvien TK (2008) Obesity and osteoarthritis inknee, hip and/or hand: An epidemiological study in the general population with 10years follow-up. BMC Musculoskelet Disord 9:5.

4. Loeser RF (2006) Molecular mechanisms of cartilage destruction: Mechanics,inflammatory mediators, and aging collide. Arthritis Rheum 54(5):1357–1360.

5. Schouten JS, van den Ouweland FA, Valkenburg HA (1992) A 12 year follow up studyin the general population on prognostic factors of cartilage loss in osteoarthritis ofthe knee. Ann Rheum Dis 51(8):932–937.

6. Katta J, Jin ZM, Ingham E, Fisher J (2008) Biotribology of articular cartilage—a reviewof the recent advances. Med Eng Phys 30(10):1349–1363.

7. Forster H, Fisher J (1996) The influence of loading time and lubricant on the friction ofarticular cartilage. Proc Inst Mech Eng H 210(2):109–119.

8. Forster H (1996)Mixed and Boundary Lubrication in Natural Synovial Joints. Doctor ofPhilosophy (University of Leeds, Leeds, United Kingdom).

9. Greene GW, et al. (2011) Adaptive mechanically controlled lubrication mechanismfound in articular joints. Proc Natl Acad Sci USA 108(13):5255–5259.

10. Greene GW, et al. (2012) Hyaluronic acid–collagen network interactions during thedynamic compression and recovery of cartilage. Soft Matter 8(38):9906–9914.

11. Guilak F, Ratcliffe A, Lane N, Rosenwasser MP, Mow VC (1994) Mechanical andbiochemical changes in the superficial zone of articular cartilage in canineexperimental osteoarthritis. J Orthop Res 12(4):474–484.

12. Ateshian GA (2009) The role of interstitial fluid pressurization in articular cartilagelubrication. J Biomech 42(9):1163–1176.

13. Berman AD, Ducker WA, Israelachvili JN (1996) Origin and characterization ofdifferent stick-slip friction mechanisms. Langmuir 12(19):4559–4563.

14. Israelachvili JN (2010) Intermolecular and Surface Forces (Academic and Elsevier,Amsterdam), 3rd Ed.

15. Drummond C, Israelachvili J, Richetti P (2003) Friction between two weakly adheringboundary lubricated surfaces in water. Phys Rev E Stat Nonlin Soft Matter Phys 67(6 Pt2):066110.

16. Balazs EA, Watson D, Duff IF, Roseman S (1967) Hyaluronic acid in synovial fluid. I.Molecular parameters of hyaluronic acid in normal and arthritis human fluids.Arthritis Rheum 10(4):357–376.

17. Dahl LB, Dahl IMS, Engström-Laurent A, Granath K (1985) Concentration andmolecular weight of sodium hyaluronate in synovial fluid from patients withrheumatoid arthritis and other arthropathies. Ann Rheum Dis 44(12):817–822.

18. Holmes MWA, Bayliss MT, Muir H (1988) Hyaluronic acid in human articular cartilage.Age-related changes in content and size. Biochem J 250(2):435–441.

19. Nagase H, Kashiwagi M (2003) Aggrecanases and cartilage matrix degradation.Arthritis Res Ther 5(2):94–103.

20. Aigner T, Kurz B, Fukui N, Sandell L (2002) Roles of chondrocytes in the pathogenesisof osteoarthritis. Curr Opin Rheumatol 14(5):578–584.

21. Sandell LJ, Aigner T (2001) Articular cartilage and changes in arthritis. Anintroduction: Cell biology of osteoarthritis. Arthritis Res 3(2):107–113.

22. McCutchen CW (1966) Boundary lubrication by synovial fluid—demonstration andpossible osmotic explanation. Fed Proc 25(3P1):1061–1068.

23. McCutchen CW (1966) Physiological lubrication. Proc Inst Mech Eng 181(Pt3J):55–62.24. Graindorge S, et al. (2006) The role of the surface amorphous layer of articular

cartilage in joint lubrication. Proc Inst Mech Eng H 220(5):597–607.25. Swann DA, et al. (1974) Role of hyaluronic acid in joint lubrication. Ann Rheum Dis

33(4):318–326.

26. Benz M, Chen NH, Israelachvili J (2004) Lubrication and wear properties of graftedpolyelectrolytes, hyaluronan and hylan, measured in the surface forces apparatus. JBiomed Mater Res A 71(1):6–15.

27. Yu J, Banquy X, Greene GW, Lowrey DD, Israelachvili JN (2012) The boundarylubrication of chemically grafted and cross-linked hyaluronic acid in phosphatebuffered saline and lipid solutions measured by the surface forces apparatus.Langmuir 28(4):2244–2250.

28. Ford IJ (1993) Roughness effect on friction for multi-asperity contact betweensurfaces. J Phys D Appl Phys 26(12):2219–2225.

29. Quaglini V, Dubini P, Ferroni D, Poggi C (2009) Influence of counterface roughness onfriction properties of engineering plastics for bearing applications. Mater Des 30(5):1650–1658.

30. Muir H (1995) The chondrocyte, architect of cartilage. Biomechanics, structure,function and molecular biology of cartilage matrix macromolecules. Bioessays 17(12):1039–1048.

31. Kampf N, Raviv U, Klein J (2004) Normal and shear forces between adsorbedand gelled layers of chitosan, a naturally occurring cationic polyelectrolyte.Macromolecules 37(3):1134–1142.

32. Raviv U, et al. (2003) Lubrication by charged polymers. Nature 425(6954):163–165.33. Lee S, Spencer ND (2008) Materials science. Sweet, hairy, soft, and slippery. Science

319(5863):575–576.34. Basalo IM, Chen FH, Hung CT, Ateshian GA (2006) Frictional response of bovine

articular cartilage under creep loading following proteoglycan digestion withchondroitinase ABC. J Biomech Eng 128(1):131–134.

35. Maeda N, Chen NH, Tirrell M, Israelachvili JN (2002) Adhesion and frictionmechanisms of polymer-on-polymer surfaces. Science 297(5580):379–382.

36. ap Gwynn I, Wade S, Ito K, Richards RG (2002) Novel aspects to the structure of rabbitarticular cartilage. Eur Cell Mat 4(Dec. 10):18–29.

37. Eyre D (2002) Collagen of articular cartilage. Arthritis Res 4(1):30–35.38. Greene GW, et al. (2010) Anisotropic dynamic changes in the pore network structure,

fluid diffusion and fluid flow in articular cartilage under compression. Biomaterials31(12):3117–3128.

39. Greene GW, et al. (2008) Changes in pore morphology and fluid transport incompressed articular cartilage and the implications for joint lubrication. Biomaterials29(33):4455–4462.

40. Lewis PR, McCutchen CW (1959) Experimental evidence for weeping lubrication inmammalian joints. Nature 184(4695):1285–1285.

41. Stolz M, et al. (2009) Early detection of aging cartilage and osteoarthritis in mice andpatient samples using atomic force microscopy. Nat Nanotechnol 4(3):186–192.

42. Venn M, Maroudas A (1977) Chemical composition and swelling of normal andosteoarthrotic femoral head cartilage. I. Chemical composition. Ann Rheum Dis 36(2):121–129.

43. Yoshizawa H, Israelachvili J (1993) Fundamental mechanisms of interfacial friction. 2.Stick-slip friction of spherical and chain molecules. J Phys Chem 97(43):11300–11313.

44. Rabinowicz E (1966) Friction and Wear of Materials (Wiley, New York), 2nd Ed.45. Han L, Dean D, Daher LA, Grodzinsky AJ, Ortiz C (2008) Cartilage aggrecan can

undergo self-adhesion. Biophys J 95(10):4862–4870.46. Han L, Grodzinsky AJ, Ortiz C (2011) Nanomechanics of the cartilage extracellular

matrix. Annual Review of Materials Research, Vol 41, Annual Review of MaterialsResearch, eds Clarke DR, Fratzl P (Annual Reviews, Palo Alto, CA), Vol 41, pp 133–168.

47. Wallace D (1985) The role of hydrophobic bonding in collagen fibril formation: Aquantitative model. Biopolymers 24(9):1705–1720.

48. Israelachvili J, et al. (2010) Recent advances in the surface forces apparatus (SFA)technique. Rep Prog Phys 73(3):036601.

49. Yu J, et al. (2011) Gecko-inspired dry adhesive for robotic applications. Adv FunctMater 21(16):3010–3018.

50. Kristiansen K, et al. (2012) Measurements of anisotropic (off-axis) friction-inducedmotion. Adv Mater (Deerfield Beach Fla) 24(38):5236–5241.

E574 | www.pnas.org/cgi/doi/10.1073/pnas.1222470110 Lee et al.

Dow

nloa

ded

by g

uest

on

June

14,

202

0